Human genome diversity data reveal that L564P is the predominant TPC2 variant and a prerequisite for the blond hair associated M484L gain-of-function effect

The endo-lysosomal two-pore channel (TPC2) has been established as an intracellular cation channel of significant physiological and pathophysiological relevance in recent years. For example, TPC2-/- mice show defects in cholesterol degradation, leading to hypercholesterinemia; TPC2 absence also results in mature-onset obesity, and a role in glucagon secretion and diabetes has been proposed. Infections with bacterial toxins or viruses e.g., cholera toxin or Ebola virus result in reduced infectivity rates in the absence of TPC2 or after pharmacological blockage, and TPC2-/- cancer cells lose their ability to migrate and metastasize efficiently. Finally, melanin production is affected by changes in hTPC2 activity, resulting in pigmentation defects and hair color variation. Here, we analyzed several publicly available genome variation data sets and identified multiple variations in the TPC2 protein in distinct human populations. Surprisingly, one variation, L564P, was found to be the predominant TPC2 isoform on a global scale. By applying endo-lysosomal patch-clamp electrophysiology, we found that L564P is a prerequisite for the previously described M484L gain-of-function effect that is associated with blond hair. Additionally, other gain-of-function variants with distinct geographical and ethnic distribution were discovered and functionally characterized. A meta-analysis of genome-wide association studies was performed, finding the polymorphisms to be associated with both distinct and overlapping traits. In sum, we present the first systematic analysis of variations in TPC2. We functionally characterized the most common variations and assessed their association with various disease traits. With TPC2 emerging as a novel drug target for the treatment of various diseases, this study provides valuable insights into ethnic and geographical distribution of TPC2 polymorphisms and their effects on channel activity.

To gain further mechanistic insight, we have introduced alternative mutations at position 564 (L564G and L564A) to assess if the hypothesis that helix-breaking amino acids are leading to the observed effect while a non-helix breaker such as A has a similar effect as L. As a result we found that only a strong helix breaker such as proline at position 564 is capable of mediating the GOF effect (see new to open state and vice versa. Based on the structure, it is evident that the L564P site is situated directly between M484L and the channel pore. We therefore assume the M484Lassociated GOF to be transduced through the L564P-encoding IIS4-S5 linker, dilating the channel pore further upon PI(3,5)P2 activation. The IIS4-S5 linker appears necessary to provide space for pore dilation and channel opening. Substituting the helix-initiating P564 with a leucine would dramatically affect this linker helix extension. We propose a model where M484L is amplifying the effect of PI(3,5)P2 activation, requiring signal transduction through L564P to result in pore dilation. A more elaborate investigation of the interplay between M484L and L564P would be however outside the scope of this paper. While being intriguing, the methods required (comparative cryo-EM of polymorphic channels and/or molecular dynamics simulations) would be both timeconsuming and resource-demanding, and outweigh the main focus of this paper, i.e. a complete electrophysiological characterization of a tremendously polymorphic endolysosomal ion channel.
(3) Most of the functional analysis is from overexpression. There is limited data from the analysis of SNP-associated endogenous TPC function from cultured fibroblasts, and it is not that convincing due to the limited number of recordings (n=3-6). In addition, the 9 donors analyzed in Fig. S6 all harbor 564P and therefore can't be used to test the effect of 564L. In order to test the effect of 564L on endogenous TPC2 function, the authors should analyze more fibroblasts and identify individuals carrying the minor allele. Given the fairly high frequency of the minor allele (~13%), this might be doable. Alternatively, even better, the authors can perhaps knock-in the 564L minor allele into one of the existing fibroblast cell line and analyze the effect.
A: TPC2 currents are generally small in fibroblasts. Nevertheless, we were able to convincingly recapitulate key overexpression results in the donor fibroblasts, in particular the discovery of novel GOF variants, K376R and G387D was confirmed (please see new Fig. S8). We have also increased the n numbers for these fibroblast lines now. Regarding human donor fibroblasts with P564L+M484L, we would like to point out again that this combination does NOT exist in humans (see Point 1). Hence, human fibroblasts with the combination P564L+M484L are unfortunately not available in any population. Regarding human donors carrying only P564L, it is clear from our analysis (please see Fig.1 world map) that this variant barely occurs in European samples. We have screened a total of 136 (!) donor samples. None of these carried the P564L variation in a homozygous manner. Screening more individuals would require again approval by the ethics committee of the LMU. Considering the time for recruitment of donors this would take several months. In addition, recruitment of indigenous people from South America or South East Asia would much more likely result in the right genotype but access to such samples seems not feasible in a reasonable time window. Likewise, a CRISPR/Cas9 approach would require several months of work from generation of the cell line to analysis by endolysosomal patch-clamp. From our perspective, the patch-clamp data in HEK293 cells have given a very clear result regarding M484L+P564L versus M484L+L564P currents and given that other data were reproducible 1:1 in the fibroblast lines, we have no reason to doubt these results.
Reviewer #2: This study explores the genetic and functional analysis of polymorphisms found in the endo-lysosomal cation channel TPC2. They used publically available dataset of human genome sequences to determine the polymorphism profiles and frequencies across the populations represented in the datasets. They then used endo-lysosomal patch-clamp analyses to show that the GOF M484L Variant only acts as a GOF variant on the 564P but not on the 564L background. The also compare activity in donor fibroblasts samples and find suggest that there are TPC2 variation patterns across the globe that alter TPC2 to become more or less active with and/or without stimulation. This is an important characterization of a gene with significant medical and global importance.

A: We thank the reviewer and appreciate his view that "this is an important characterization of a gene with significant medical and global importance."
Comments: Page 9. The authors state that there is a higher variation frequency for TPC2 than for similar channels. While it is true that there more polymorphisms found for TPC2 than for other channels I am not sure if this is significant. Did they control for exon size and look at noncoding/intronic region variation of these genes to make this statement. If not maybe just state the frequencies without a qualifier as it being higher variation. Maybe say the most variation was seen in.
A: We have now calculated the percentage of polymorphisms (see Fig. F below; left: all SNPs; right: only homozygous SNPs) for TRPML1, 2, 3 and TPC1, 2, relative to the coding sequence length for each gene. We omitted intron lengths from this analysis as intron variations were not a focus of this study.
In transfection and endogenous patch clamp studies, how did authors control for equal protein levels and correction cellular locations of TCP2 variant?
A: By measuring current densities = current in pA divided by capacitance in pF we can normalize to vesicle size and thus control indirectly for the average number of channels measured. All channels are localized to lysosomes, otherwise it would not have been possible to patch-clamp them (if e.g. mislocalized to ER or cytosol). We have also confirmed this now by confocal image analysis (see new Fig. S6). In addition, we have now performed a WB analysis to assess expression levels (see new Fig. S6). We found no significant differences in expression, rather GOF mutants appeared to show a trend towards lower expression. This is in line with previously published results (Chao et al., PNAS, 2017 Suppl. Fig. 1C and D). Importantly, no significant differences were found for M484L+L564P versus M484L+P564L. It can thus be ruled out that expression or mislocalization are the underlying causes for the observed GOF effects or the differences between M484L+L564P and M484L+P564L.